cancel
Showing results for 
Search instead for 
Did you mean: 

Is there a max number of policies (either hard value or best practice)?

zmlat
Level 4

Hi,

I'm in the midst of planning a refresh of our backup environment, and go to version 8.1 on Linux servers. Management is asking for a particular setup that would result in 1000s of policies (at least 1 per backup client). As such I was wondering if there was a max # of policies, or a best practice that might justify keeping that # under a certain amount. While it would be "ugly" to set a single policy per backup client, it would align with some internal automation tools to add/remove backup clients. Also wondering if this methodology aligns better with NetWorker (since the folks pushing for this use NWK for backups).

Thanks

6 REPLIES 6

Thanks for the reply, but the link doesn't really answer my question. Having worked with NBU for a long time, I prefer to have as little policies as possible, typically one per platform (unix or windows), per environment (prod, dev, uat, etc...). Easier to manage. This proposal to shift to one policy per filesystem goes against the grain. So I'm wondering if there is a technical reason to avoid this methodology.

I'll just say for my environment, the following is done.

1 Policy per environment (dev, qa, prod) per OS per LOB (line of business)
So 1 windows policy for windows, prod, lob1
1 windows policy for windows, prod, lob2
1 standard policy for linx, prod lob1
etc etc

Also currently have DB/Archive Log backups using 4-5 policies per DB that utilize BRTools due to security restrictions atm, we don't use Oracle Intelligent policies.

We do the same thing for VM's using tagging in vSphere.

Currently have over 800 policies with no issues.

Genericus
Moderator
Moderator
   VIP   

My environment is a lot like krutons - grouped by OS/environment/LOB.

I name my policies with a standard PREFIX based on OS - UNIX/LINUX/WIN/RMAN - this tells OPS who to call.

I end the policy name with a standard SUFFIX based on environment TEST/DEV/QA/PROD - this tells OPS what to do - page on PROD, email on DEV, etc. SInce I standardized my policy names, I was able to automate the backup_exit_notify script to email OPS and the specifc ENV/LOB group on failures. A big help.

There are specific valid reasons for grouping clients/policies, and there are reasons to separate them. I know I see issues where a client with their own policy that backs up once a year, ALWAYS shows up as a problem because it has not backed up within the last 30 days...

There are reasons not to mix calendar and frequency schedules.  There are reasons to combine daily/weekly/monthly schedules within one policy. 

My suggestion - PLAN, consider how your plan will adapt to changes - what if you add a new environment, LOB etc?

PLAN, peer review, PLAN then build.  good luck!

 

BTW - I have 386 policies and a compressed catalog just under 1TB. 1 master, 25 media servers and 18 physical tape drives. LOL, and over 900 virtual ones!

NetBackup 9.1.0.1 on Solaris 11, writing to Data Domain 9800 7.7.4.0
duplicating via SLP to LTO5 & LTO8 in SL8500 via ACSLS

Great advice Genericus.

I cannot stress how important it is to plan your environment and deployment. It is not easy to overhaul a change to your entire environment and sometimes it just isn't possible.

 

Mike_Gavrilov
Moderator
Moderator
Partner    VIP    Accredited Certified

NetBackup doesn't have limitation on policies but only one backup job can be launched per second, regardless of the resources available. As a result, a single NetBackup domain can launch a maximum of 86,400 and Veritas recommends 30 000 job for a single domain. 

Genericus
Moderator
Moderator
   VIP   

Kudo to Mike for the actual answer - LOL, how many seconds in a day? 86,400, so it can only start that many in one day.

So keep it under 90,000 please!

I have 376 policies, and that is too many, but I keep deactivated ones around ( for too long ) .

 

Hey Veritas  - how about populating clients for restores based on the /db/images directories instead of the clients in policies? I could get rid of some right there. 

NetBackup 9.1.0.1 on Solaris 11, writing to Data Domain 9800 7.7.4.0
duplicating via SLP to LTO5 & LTO8 in SL8500 via ACSLS